Bitcoin Forum

Somehow we worry about going to 2MB/block for fear of 'bloat', yet this is how much memory costs:

https://image-store.slidesharecdn.com/13386f23-5dbe-41e7-a99d-07583b215b66-original.png

Today, you can buy enough memory for the entire blockchain for a few dollars.  

I think the larger problem is network bandwidth rather than hard device capacity. Even with high speed Internet connection (> 20 Mbit/sec) it can take a couple of days to download the block chain from scratch.

Yes, that is exactly what Chines miners are worring about. The increasing block size may impact their mining facility and the rate of mining reward.

lol.  You only have to download the chain one time - forever. 

People sit in their living room all over the planet streaming movies every night and you worry about 2MB every ten minutes?  Clearly you failed your math A levels. 

OP, how much bandwidth has your full node used in the last month? What's your uptime like? How many connections do you normally have?

lol of course, because of the moore's law. and we have less silicon needed to produce the same amount of memory.
moreover this is based on HDD but not yet recording what SSD just did.

Yeah, most of the time the processing power + storage are actually increasing by more than Moore's law. I

t's going to get cheaper with the years coming.

This is true. However many people have high pings and difficulty getting a few kb through censorship walls. Bandwidth is an issue for some. But is it an issue for 2mb blocks? That I guess we don't know, unless there are chinese miners around :)

not to mention those same people are willing to download easily from any torrent 60 gb of games, without thinking, so there is must something wrong in their brain function i believe....

Memory is pretty cheap these days you can get loads of memory for a god price. Technology is getting better and you know have more memory on a smaller platform so it is only normal for memory to get cheaper.

So China has it's great firewall with very limited bandwith, but they also have the cheapest (and dirtiest!) electricity. The cheap electricity made them the main miners, pushing all competition off the market, and since they all share the same interests, they have the power to keep blocks small.
I see a chicken-egg problem, which limits the growth of Bitcoin.

What is this?
Another stupid thread and attempt to scream for bigger blocks??
The roadmap is out!And it's a good one.This had been discussed in a hundred threads now.
Getting tired of this.

i think the issue with the bigger blocks isnt the hard drive space..
its the fact that right now you have to include a fee to send bitcoin.. if you dont it will surely get rejected and not sent..
but once the blocks are made bigger there will be so much empty space that even zero fee transactions will get included.. and when those blocks get filled people will start screaming for even larger blocks..

this is a problem, because the miners are supposed to get the fees.. and if you make them less needed.. they will get less..

the size even, it just shows how smart human minds can develop and achieve

I can't be the only one that remembers the time when user to user data transfers would mostly happen offline because home internet was unreliable and bandwidth was expensive. This is still the case for many countries, and home internet isn't exactly cheap for most of the world. Hard drive space isn't the only issue.

The blockchain size isn't too much, we download many things online and when you think you download all the blockchain for less than 100GB it isn't too much. I'm pro 2MB/block.

Here we go again, threads that are a result of inadequate research. 2 MB block size limit:

• Increase storage requirement (the blockchain is already huge(
• Increased network requirement
• Attack vector at 2 MB (validation that takes longer than 10 minutes)
• 2x TX capacity

Segwit gives you almost the same amount of TX capacity while avoiding the attack vector, scaling down the validation from O(n^2) to O(n) making it linear. It also comes with other benefits and fixes required for LN.

And electricity costs are an issue for others. If mining becomes slightly less centralized in China, I won't lose any sleep.


Segwit, or rather the promise of segwit in another Two Weeks^TM year, give me exactly the same amount as NOTHING.

And it will only keep falling in the coming years  ;D

Irony alert lol. Cheap RAM is an argument for Lightning channels, not Gavin Andresen BloatCoin. Duh, OP. Duh.

Exactly.

The BloatCoiners love to make fancy charts. They are experts at linear extrapolation with very limited real data or by showing data irrelevant to the issues that are being debated. Remember that alarming graph that showed hitting the block limit with all painted red and drama? Guess what: Nothing happened.

To repeat it here for the 100th time: Memory is not the bottleneck. The bottleneck is the network.

Also in general, assuming that capacity increases of the past decades can be extrapolated into the future without any decrease in the growth rate is science fiction. There are physical limits in miniaturization. We are already very close to these limits.

ya.ya.yo!

No you don't just download the chain and be done with it. A full node does not only download blocks - it also receives and relays unconfirmed transactions and sends out blocks to other SPV clients.

Your node could easily send out terabytes of data in 1 month if you don't restrict it.

This node has been up for only 6 days and has already sent out 144 GB.

https://i.imgur.com/X7GT76b.png

Just another example of a person who clearly has no idea what he's talking about. No idea how I missed this idiotic post on my first visit to this thread. You obviously don't even know how nodes work. why nodes spend this much bandwidth. Seems like the original statement was correct (more or less) and I had misinterpreted it.

Correct. Even nodes with lower upload connections and total number of connections send out a lot of data. In April my node spent ~150GB of data.

---

---

Update: Removed initial statement as it was a misinterpretation.

I know it is such an amazing thing

lol!  Fuck you.  I definitely know what I am talking about.  144GB is NOTHING!!  You must be a very cheap son of a bitch to complain about this bandwidth since one can buy unlimited for about $100/yr.  Your provider is in Zimbabwei so you pay more.  

You ONLY have to download the chain once.  True, a full node relays shit all day long - but my original claim that you only have to download the chain once is correct.  So now who is the fuck wad that doesn't know what they are talking about?   That would be you.  Moron.

Your node sent out 150GB?  That has nothing to do with memory asshole.  Read the title of the thread.  Now who doesn't know anything.  Bandwidth has nothing to do with memory.  My claim is the memory is cheap.  Bandwidth is also cheap but that is not what the post is about.  By the way, you only have to download the blockchain once.  Get it dumb-ass?  Just because you sent out 150GB last month, your memory requirements didn't go up.  that is your bandwidth requirements.  How many times did you download the entire blockchain last month?  ZERO!  That's right, because my first claim was in fact correct.  You only have to download the blockchain once.  So a full node has some bandwidth issues - OK - great topic go start your own thread.  That has nothing to do with the fact that a HUGE blockchain can easily be accommodated on a bunch of cheap memory.

Lauda - you are a fuck wad.  When you start telling people they don't know what they are talking about - and then you do change the topic -  you are total bullshit.  Stick with the topic - memory is cheap, the blockchain can get huge and it doesn't cost much to accommodate it.  If you have having trouble with bandwidth - get a better subscription.  But don't say I don't know what I am talking about because I know the difference between memory and bandwidth.  Fuck off.

Yeah that is definitely the bigger problem. Where I live, the speed of internet is very very slow. It is like 10mbit/sec. I really hate it and I hope the internet companies do something about it.

144GB is nothing indeed (depending where you are), however you've missed the "in 6 days" part.

Please link me to the ISP with an unlimited plan with decent speeds that is <$10 a month. Otherwise this is just hyperbolic nonsense.

The initial download is not what the problem is nor the problem with your statement:
Assuming that only 2MB will be spent every 10 minutes is very wrong.

---

---

There's really no need to resort to childish behavior because I've called you out on this mistake (note: Nobody said that you don't know the difference between memory and bandwidth).

People advocating for contentious HF's don't really care about that apparently. I can also picture memory being more expensive in those areas, am I right (storage can be quite cheap but only depending on where you're from)?

---

---

How? I was told that it was only that expensive in Zimbabwei ???

---

---

Update: Several updates to post; more are probably coming.

Actually that particular node is hosted in an Azure VM in their Japan East datacenter and costs me $120-$180 a month to run, depending mostly on how much bandwidth I use that month.

That was my assumption for movies - it was not an estimate of a full node bandwidth.  

Bandwidth is still cheap.  Your Azure package includes LOTS of other services - in addition to the bandwidth.  You are paying this price for the package of services.  You can have VERY cheap bandwidth - if that is all you need.  Your fancy Azure package is what is expensive - not the bandwidth.  

Actually, there is a need to resort to childish behaviour.  When you start your argument with 'doesn't know what he is talking about' - you've already abandon a logical response and started a personal attack.  You therefore are the childish asshole and need arises to address childish assholes with stuff they understand well.  I am just trying to get on your level my friend.

I am saying you don't know the difference between memory and bandwidth.  The post is about cheap memory.  Look at the chart.  Bandwidth is another issue - if you live in Zimbabwei.  Bandwidth is very cheap everywhere else.  Don't point to your Gucci Azure package - that fancy VM and other services (i.e. 100% uptime) is what really cost money - not the bandwidth.  

Well, then it is technically correct (similar (but not the same) with a non-node full client).

I'm not using Azure. This node is located in a remote location with a normal domestic connection (albeit, almost the top tier that is available). The last time that I've talked to a higher ISP representative (which was 2-3 years ago) they've told me that upload bandwidth is quite expensive and this is why their upload speed was much lower compared to the download one.

---

---

That's not the intend behind the statement; I've just 'written down' the impression that I had. I've 'pulled it back' as it was a misinterpretation; there was no need to be hyperbolic about it.

That would be a ridiculous statement (as my background is IT). I think that you're confusing Soros Shorts's node (which is on Azure) and mine (residential connection).

---

---

Back to memory: It does not cost much (even though we should not generalize), especially in 1st world countries (e.g. USA). It might be a bigger sum for people located in 3rd world countries. My node is currently on a 250 GB drive which I've picked up for $20 and there's plenty of space left (70+ GB).

Glad to see you've admitted the errors of your ways and now agree with the original post.  It takes a big person to admit they were wrong.  Next time - don't launch your disagreement with a personal attack.  Sorry your upload bandwidth prices (from 2 years ago) are bad.  However, on a decent provider they remain cheap - even in view of the high load that full nodes demand.  Now get out of Zimbabwei if you want to run a node.  Maybe move to a decent country like Germany or Holland. 

Yes ... because of the network usage ...


How does RAM help my bandwidth cost?

I see this as a virtue, and as a rational person (I'm trying) it makes sense to admit to wrong and try to improve.


I have not listed any prices? I have just elaborated as to why the internet speed in some countries are (e.g. 50/5) as I've asked.

I don't live in such a country. My node works fine, even though it is not as useful as Soros Shorts's node (mine is not able to provide that much data).

RAM is not the only type of memory. OP is talking about storage.

Its not memory were worried about its the bandwidth, latency and orphan rates.  

How do you restrict the amount of data a node can process?

I want to set up a node but the location I live in has poor internet connectivity. The only way I can set up a node is through a server hosting provider. There's a limit on how much data I can afford to pay for each month, so I want to put a cap on my monthly data.

25 years ago when circuit feature size was 1 micron - we were saying the exact same thing.  I really don't know how anyone is doing 14um.  That seems totally impossible.  But - whatever, they are doing it.  

I don't think memory capacity will continue to go down like the chart for the last 20 years.  It will go down however.  Now, memory is so fucking cheap it is shocking.  I once paid $100 / Megabyte (RAM).  Now when I buy 128GB flash for $30 I laugh my head off.  

As for bandwidth - the arguments here are hard to accept.  Some large portion of the Internet bandwidth is streaming porn.  I am pretty sure the purpose of Bitcoin exceeds the purpose of watching another very large black man vigorously pound some poor teenage pussy into oblivion.  We already have very significant bandwidth availability.  Maybe some of it needs to be redirected - and it will as soon as the 'importance' of bitcoin transactions exceeds the 'importance' of having a look at sex sites.  

Yeah that's not strange why it's getting cheaper and cheaper because new technology is coming.
There are allot of things that changed in a short period not only the memory went smaller and larger amount of GB or TB.
But the whole computer has grown very strong.

It sure is nice to have you two geniuses to help out.

i think better solution for this could be cd or pendrive with the copy of ledger which can be released on yearly basis. So that the one who like to host own wallet can download remaining few blocks within few hours.

The problem arises when you reach the size of 1 atom.

Again, it comes down to where the user is from. While $30 might seem like a small amount in the US, it is a big amount in remote places in the world. It comes down to whether and by how much you want to restrict node usage. Do you want nodes only to be run by people in 1st world countries?

Nothing prevents you from selling one, but it will get outdated pretty quickly.

Memory will only gets cheaper and cheaper because there are so many new things coming up.
Why are you not happy about it is better then paying the biggest price.

You didn't read a single post in the thread, did you (e.g. "why are you not happy about this")? Designing software with an optimistic view on growth rates ends up with disaster.

The good news is we have a long way to go.  14nm is 140 angstrom.  An atom is only about 1 angstrom in diameter.  So, there remains tons of room for size reduction before we get to the atom.  Looks like memory is going to be even cheaper. 

Just curious, how many people who consider $30 a large sum of money currently run nodes? How would it matter if Angola got a few? Describe in detail plz.

http://s32.postimg.org/o9dmglyg5/Capture.png

exactly, cheaper and with a lot more of capacity, so in the end, there will be no excuse to run a node, yes? well now we just need that our isp cheapen prices also with better bandwith and bitcoin nodes will grow up like it deserve ;)

Depending on who you quote you have: (so yes)There is still room for improvement indeed. However, I don't have to tell you that it becomes more and more difficult to manufacture them (as you probably know this).

You make it sound like switching to that would be easy.

So you want to restrict nodes to only developed places (1st world countries)? So the end result is even further centralizing Bitcoin than it already is, correct? The reasoning is that you hope that: more capacity = more users = higher prices, right?

If X amount of people consider Y to be a large sum of money and Y is required to run a node, then obviously people from X would not be running a node in the first place. I'm not sure why you posted a map of the nodes though. As for that $30 example, take a look at any country where the minimum daily wage is <$1.

---

---

No. There's no financial incentive to run one. While it might be true that: more users == more nodes, we would be just in the area of speculation.

So people living on <$1 a day will never run a node, not even if we cut the blocksize in half.
Further, those living on <$1 a day have less interest (or use for) bitcoins than they do for 20% discount on Bentleys or Panamanian shell corporations.
These people are not a part of the Bitcoin ecosystem, certainly not as far as nodes are concerned. So why are we even talking about them?

You're the one who started talking about them, not me. I was making an example of subjective view on money, the figure $30 is arbitrary. Re-read my initial statement.

My point either... ::)
if they cant afford 30$, they probably dont have internet or even a computer, so... whats the point? they wont be using bitcoin anyway.

So you're not talking about poor people in third world countries?
What are you trying to say?

Which was followed up by questions to OP and his views. I don't see myself talking about people that are not part of the ecosystem, nor people for which $30 is a lot in particular (aside from replying to you). Focusing on a single sentence or figure will lead up to a misinterpretation of the post.

---

---

This is becoming off-topic.

Well... that i dont agree.
Wouldn't be centralization because a lot of places can afford that so would still be decentralized.

right now the main problem with bitcoin centralization seems to be energy for mining, where is way more profitable to do in china... so all big farms are there.

bandwidth or memory ?

OP is is addressing the costs of "bloating" the blockchain with 2MB blocks, suggesting that cost increase would be trivial.
You, if I understand you correctly, counter by saying that trivial sums are not trivial for everyone the world over, and higher costs of running a node may adversely affect decentralization, limiting nodes to first world countries.
You punctuate your point by asking OP: "Do you want nodes only to be run by people in 1st world countries?"

Is this reading not accurate?

More-or-less, yes. However, I'd rather that you PM me about my posts than us further spiral into off-topic discussions.

You can't really disagree with it. I never said that the network would be centralized, I said "more centralized" which is quite different.

I don't understand where people get this impression from? Bandwidth is not that cheap.

Dunno, probably pictures like this:
https://cdn0.vox-cdn.com/thumbor/FcEobBZEWJTs4Or3aJ2J7rnHr5Q=/cdn0.vox-cdn.com/uploads/chorus_asset/file/673658/bandwidth_cost.0.png

Sure, "cheap" needs to be qualified, and people living on <$1 a day most likely consider current prices prohibitively expensive, but until we have socialized internet access, this can't be helped. The point is prices are falling, so the one meg that cost us X a few years ago now costs us X/2.

Where do you actually live Lauda?  Bandwidth is fucking nearly free.  Unlimited bandwidth for $10/month at many providers.  Of course 'unlimited' is nonsense, but it is so fucking cheap they don't bother to measure it any longer.  Most hosting providers stopped bandwidth limits a long time ago.  Even heavy bandwidth users aren't over loading the system.  Only Netflix and Brazzers is hard to handle. Bitcion nodes are FUCKING TINY in comparison.  



Exactly.  It is now too cheap to even meter.  So 'unlimited' can be purchase on the shittest hosting plans.

Memory is getting pretty cheap these days because we are getting farther which technology. It is now possible to store memory on smaller space. I wonder how the memory is going to be in the future when we are even further.

That is true, yes.

That does not matter. I'm not making these statements because I would run into bandwidth problems (my node is on a unlimited plan, and so am I). I'm not exactly sure how much the internet package costs for my node.

I'm still waiting for examples of these providers (at decent speeds!). Are you talking about residential internet or servers?

I'd like to see data and a comparison chart (maybe someone could create it). However, if you're saying that 140 GB in 6 days is tiny in comparison to Netflix usage (of a single person?) in that time period then that is strange.

Can you explain how you're getting 140 GB per 6 days (~24 GB/day)? Serious question.

Here you go:
-topic questions, etc.).

I added some color for you. Misconfiguring a node with some ridiculous maxconnections (running it wide-open) is possible, and even a nice thing to do if you have unlimited bandwidth. It's also nice to tip your waiter $500 when you go out for lunch.
Doing so is in no way a prerequisite, for running a node or eating out.
Hope this helps :)

Here is some math for you, I really like math.

2 MB per 10 mins = 12 mb/hr

12 mb/hr * 24 hours = 288 mb/day
288 mb/day * 365 days = 105,120 mb

105,120 mb = 105.12 gb/yr

The math above should be pretty simple to understand, just basic multiplication with a small case of division at the end to turn mb into gb.

Storage shouldn't be an issue for most computers (right now anyway) as the standard laptop has 500 gb and a desktop usually has 500 gb - 1 tb of memory.

But, lets say that the 2 mb blocks are added and 2 years go by and I discover BTC. I download the core wallet and start to sync. The blockchain is already around 68 gb so we can do some more math here yay.

105.12 * 2 + 68 = 278.24 gb I would have to download. Now lets take into consideration the price of internet right now.

I pay $94/month for 500 gb of bandwidth. Roughly $0.19 per gb.

Now lets add up that cost with the size of the blockchain a couple years from now. Granted this won't be accurate as internet prices in my area have been increasing as the years go by but it will be close enough.

$0.19 * 278.84 = $52.98

It would cost me over $50 to download the btc wallet which is ridiculous. On top of that it would use 3/5 of my laptop hdd to download it. Now storage becomes a problem. Not only is it expensive to download one program but it will use up almost all of my available storage.

This isn't even taking power usage into factor. It is going to take some electricity to get that massive blockchain synced so will only increase the cost more.

Glancing through some of the other posts on this thread I saw unlimited bandwidth mentioned a few times. Here's some news for those that think unlimited will last forever. It won't. It costs ISP's too much money to let their customers just use what they want.

My ISP is one of the largest in the US (Suddenlink) and we used to have unlimited internet. Well, not anymore we don't and I foresee unlimited bandwidth going away for computer users in the near future.

And how much would it cost you to download the wallet if the blocks remain @1MB? Could you help me out with the math?

You are starting to lead a unhealthy discussion. The only concrete data provided was either his or mine (numbers), and there is nothing more that could be referenced from within the thread. Assuming that his node is misconfigured just because it spent a lot of data within 6 days is foolish at best. It doesn't have to be his node, let's look at Shorena's node:

https://i.imgur.com/1wluEWq.png
5 days = 247.21 GB. Case closed.

Obviously 1/2 the cost of the proposed which is only $26.. Use your brain man don't rely on other humans to guide you step by step through life. I learned this math in elementary school 20 years ago..

I added some color for you. Misconfiguring a node with some ridiculous maxconnections (running it wide-open) is possible, and even a nice thing to do if you have unlimited bandwidth. It's also nice to tip your waiter $500 when you go out for lunch.
Doing so is in no way a prerequisite, for running a node or eating out.
Hope this helps :)

Edit: Still waiting for halp :(

By what comparison is that a lot of money? All the things you listed we already have, and the price of the said things in only falling down with advancement
of the technology that's happening litleraly every day forward. imho, arguments for consumption and prices of maintaining nodes are small enough and
even larger blocks pose no real expense.

Allowing more users == misconfiguration for you? That is not correct and the data that I've provided fits (50-80 connections). The analogy does not fit either.

This depends on min fee parameters.

I think you didn't memory, but megabytes of space?
Anyway, the main problem with raising the block size is that you inevitable centralize nodes because the countries that need nodes the most are usually countries with impoverished technology with shitty economical situation venezuela. It's important that nodes are run all over the planet, not only the total node count. I hope that eventually we can hard fork for a higher MB block size, but it is not a priority at all.

Do you honestly not understand what I'm telling you? Tipping your waiter $500 on a $20 meal is also, technically, not a mistake (misconfiguration), but only if you understand that you're leaving an unusually large tip & can afford to do so.

For those of us living in third world countries, on <$1 a day, OTOH, leaving a $500 tip would be a mistake.
In all honesty tho, I doubt they go out to fancy restaurants anyway. So the entire issue is as moot as the cost of running the node which they do not run anyway (because they can't afford to use bitcoin).

What are you talking about? Running a node with the default configuration will easily use this much bandwidth.

Still doesn't fit. You've chosen to ignore the data again:

Apparently stock configuration means that your node is misconfigured.  :D

I don't have time to deal with trolls or people who are stupid trying to act smart. If you want to actually have a discussion then I would suggest reading and not posting nonsense.

It would have taken you far less time to say "It would cost me a ridiculous sum of money *regardless of blocksize*, because I pay $94/month for 500 gb of bandwidth." I would commiserate, and say something like "I hope that's a tv/phone/internet deal, or you live in some God-forsaken wilderness, 'cos otherwise you're paying way too much."

Had you chosen to elaborate, you could have also mentioned that you can't into math, because confusing 2MB blocksize limit with every block being 2MB, something that has, roughly, zero chance of being the case.

At that point I would have taken pity on you, and helped you with your math.

Getting ~3GB/day here : http://213.165.91.169/

That is unusually low. Are you limiting your connection or something else somehow? What are your internet speeds? At this very moment my node has about 4-5 times more unconfirmed TX than your node.

Memory in the past wasn't cheap but because people can make smaller chips with more memory on it. The
old ones will become very cheap. And after a year the newer ones gets cheaper because there is a new
generation.

In the past memory was pretty expensive but know with technology improving it is allot cheaper then before. They can now store more memory in a smaller platform which is ideal in most situations. 

Technology moves forward in a very quick fashion.

I remember buying a 128MB memory card was very expensive which made me stick to 64MB cards for quite a while. ;D

Now you can buy MicroSD cards with 1TB capacity and even higher if you so wish.

So much memory in such a small piece of plastic. It's mind blowing.

This post makes little sense and does not necessarily contain the right information. The old generation of memory did not become "very cheap" (e.g. DDR2 2GB costs more than a DDR3 2GB stick). Unless you're talking about storage, then that might be a different case.

It really isn't about the space required to store memory when it comes to Bitcoin.

The only thing that is mind blowing is that there are some fools that are willing to pay insane premiums to get those cards. I've managed to find a 512GB card and it costs a ridiculous ~350 euros.

---

---

Did some small calculation: If we assume a 2 MB block size limit and that 100% of the blocks are full (which is unreasonable) we have:
2MB * 6 blocks per hour * 24 hours a day * 365 days a year = 105.1 GB. A more reasonable view (75%) would result in the blockchain size increasing at around 78 GB's per year. This doesn't seem as much, but I'd hate downloading that from scratch (combined with the size today).
The price per gigabyte in 2015 is 0.03$ (according to this (http://www.mkomo.com/cost-per-gigabyte-update)) which equals to ~2.4$ for that year. Which does not seem much at all (if we assume that these prices are available everywhere).

Most memories are cheap now a days

exactly, 2MB every 10 minutes is basically nothing.

that's about 416 Bytes per second (~3Kb)

even a dialup 56k modem could easily deal with that.

Did you forget that blocks have to be validated? You can't just do the calculations like that and assume that the network would be healthy if it took 10 minutes to receive each block (assuming regular time between blocks). There's even that potential attack vector at 2 MB where the validation could take more than 10 minutes. However, this would go a bit away from the topic(?).

Memory was expesive but know you can buy more memory for less money. Technology is going so fast these days some people even say that is goes to fast.

memory these days are cheap because there are so many sorts of them. They are getting smaller and smaller each year but the memory gets bigger and bigger how is that even possible. I bought a 2tb memory stick about 4 years ago and yesterday i bought a new one with the amount of 4tb the one with 4tb was half of the size of the one i bought 4 years ago that's so strange.

That is not the cause. The principle cause is miniaturisation of the process that manufactures the memory types, not the range of types itself.


Again, miniaturisation is the reason for your wonder.

The post is about cheap memory. Bandwidth is another issue. Bandwidth is very cheap also. The only issue is with the Network Infrastructure, the protocols used and handshaking between different types of network and so on.  
Look at the chart again. It is related about Bandwidth and not about the cheap memory. :D

i guess its about the difficulty to mine as years goes by the deeper and deeper it would be. just like gold as the time goes by it is harder to find and its
price is increasing because of the difficulty same in bitcoin. So the mining site will provide new technology to cope up in mining bitcoins and it will be m
more expensive to us to buy more powers.  ;D

Correct.

They're correlated in this case.

The definition of 'cheap' is subjective and internet speeds (and provided plans) have a huge variation based on regions. This is why the global average speed is quite low.

This post does not have any relevance to the original post. In other words, signature spam.